What can researchers do on the basis of knowing a trait’s heritability if the genetic and environmental factors underlying the observed trait are heterogeneous, or if the method of data analysis does not allow researchers to rule out the possibility of underlying heterogeneity?  What steps and conditions are needed for researchers to bridge or circumvent the knowledge that underlying factors may be heterogeneous?  There seem to be six directions that researchers might pursue:

a. Undertake research to identify the specific, measurable genetic and environmental factors without reference to the trait’s heritability or the other fractions of the total variance (e.g., Moffitt et al. 2005, Davey Smith and Ebrahim 2007, Khoury et al. 2007).  Discussion of this direction of research takes us beyond heritability studies.

b. Use high heritability to guide molecular research to identify the specific genetic factors involved.  There may be traits for which the underlying factors are not heterogeneous.  These might be worth finding even if researchers do not know in advance the proportion of fruitful investigations compared with those confounded by the underlying heterogeneity.   The search here is not for high penetrance major genes; these can be detected through examination of family trees; heritability analysis need not be involved.  Rather, researchers need to find traits in which many underlying genetic factors each of small influence turn out to be similar for all individuals who show the same value for the trait within some defined population.

c.  Restrict attention to within a set of relatives.  Even if the underlying factors are not yet known, high heritability still means that if one twin develops the trait (e.g., type 1 diabetes), the other twin is more likely to as well.  This information might stimulate the second twin to take measures to reduce the health impact if and when the disease starts to appear.  However, notice that this scenario assumes that the timing of getting the condition differs from the first twin to the second.  Researchers might well ask: What factors influence the timing?  How changeable are these?  How much reduction in risk comes from changing them?  To address these issues researchers have to identify the genetic and environmental factors involved in the development of the trait and to secure larger sample sizes than any single set of relatives allows.  The question then arises whether the results can be extrapolated from one set of relatives to others.  The possibility of underlying heterogeneity has not, therefore, gone away as an issue.  The answer to the question of extrapolation is an empirical one; there is a risk, as before, that the proportion of fruitful investigations will be low compared to those confounded by factors not extrapolating well from the initial set of relatives.

d. Put aside the search for measurable factors.  Instead, focus on heritability as a fraction of the variation among measurements.  This focus is useful in agricultural and laboratory breeding.  If the actual advance under selective breeding is less than predicted, one source of the discrepancy might be the underlying heterogeneity of genetic factors and their reassortment through mating.  Again, this matters little for breeders because they can always compensate for discrepancies: they discard the undesired offspring, breed the desired ones, and continue.  Of course, selective breeding is not an acceptable option for humans.  What is left is the intuition that genetic factors have a larger influence than environmental factors for high heritability traits.  This is problematic (Taylor 2010, p. 13ff), even more so when researchers consider models that allow for heterogeneous factors to underlie the trait.

e. Reduce the possibility of underlying heterogeneity by restricting the range of varieties or locations.  Agricultural researchers can reduce the possibility of underlying heterogeneity by restricting the range of locations in which a variety is raised or grown.  They can also control environmental conditions, such as, for animals, the regimes of feeding and husbandry or, for plants, the application of fertilizer and irrigated water.  Agricultural breeders can also produce inbred lines and thereby eliminate the heterogeneity of genetic factors that exists within outbred varieties.  However, to envisage taking action on the basis of research conducted under restrictive conditions is to presume that the restrictive conditions can be replicated.  This presumption is most apparent when plant breeders recommend varieties to be grown only in defined regions and under prescribed techniques of cultivation, or when animal breeders specify the optimal feeding and husbandry for each variety.  In the study of human traits, however, it is not feasible to control the full range of relevant environmental conditions or to breed for genetic uniformity.  It may be possible to restrict the locations included in a human study (e.g., to include only families of low socioeconomic status; Turkheimer et al. 2003).  The heritability estimates would be reliable to the extent that these restrictions were replicated in subsequent research or policy.  That is, the research could be applied even though the environmental factors underlying those locations had not been identified.

f. Reduce the possibility of underlying heterogeneity by grouping varieties that are similar in responses across locations. (Note that when analyzing measurements from studies of human twins because such studies have only two replicates [twins] in one or at most two locations [families], so this is not a feasible direction by which research on human variation can bridge or circumvent the knowledge that underlying factors may be heterogeneous.)  In agricultural trials, where a number of varieties or animals or plants can be raised or grown in multiple replicates in many locations, varieties can be grouped by similarity in responses across all locations (using techniques of cluster analysis; Byth et al. 1976).  (Similarly, locations can be grouped by similarity in responses elicited from varieties grown across those locations.)  Varieties in any resulting group tend to be above average for a location in the same locations and below average in the same location (Taylor 2010).  The wider the range of locations in the measurements on which the grouping is based, the more likely it is that the ups and downs shared by varieties in a group are produced by the same conjunctions of underlying measurable factors.  This gives researchers more license to discount the possibility of underlying heterogeneity within a group.  If the underlying factors are assumed to be homogeneous within each of the groups, researchers can hypothesize about the group averages—about what factors in the locations elicited basically the same response from varieties in a particular variety group that distinguishes them from other groups.  (It should be noted that knowledge from sources other than the data analysis is always needed to help researchers generate any hypotheses about genetic and environmental factors.)

In summary, unless you can think of directions of research other than the six above, there is very little that researchers on human variation can do that is reliable on the basis of knowing a trait’s heritability if the genetic and environmental factors underlying the observed trait are heterogeneous.  Agricultural researchers can do more because they have greater control of their varieties and conditions in test locations.

Adapted from Taylor (2010) and Nature-Nurture? No… A Short, but Expanding Guide to Variation and Heredity (work in progress)

References

Byth, D.E., Eisemann, R.L. and DeLacy, I.H.: 1976, Two-Way Pattern Analysis of a Large Data Set to Evaluate Genotypic Adaptation. Heredity 37(2), 215-230.

Davey-Smith, G. and Ebrahim, S.: 2007, Mendelian randomization: Genetic variants as instruments for strengthening causal influences in observational studies, in Weinstein, M., Vaupel, J. W., Wachter, K.W. (eds.) Biosocial Surveys. Washington, DC, National Academies Press, pp. 336-366.

Khoury, M.J., Little, J., Gwinn, M. and Ioannidis, J.P.: 2007, On the Synthesis and Interpretation of Consistent but Weak Gene-Disease Associations in the Era of Genome-Wide Association Studies. International Journal of Epidemiology, 36, 439-445.

Moffitt, T.E., Caspi, A. and Rutter, M.: 2005, Strategy for Investigating Interactions between Measured Genes and Measured Environments. Archives of General Psychiatry, 62(5), 473-481.

Taylor, P. “Three puzzles and eight gaps: What heritability studies and critical commentaries have not paid enough attention to,” Biology & Philosophy, 25:1-31, 2010

Turkheimer, E., Haley, A., Waldron, M., D’Onofrio, B. and Gottesman, I.I.: 2003, Socioeconomic Status Modifies Heritability of IQ in Young Children, Psychological Science 16(6), 623-628